The wood anatomy of the polyphyletic Icacinaceae sl, and their relationships within asterids

525
Lens & al. • Wood anatomy of IcacinaceaeTAXON 57 (2) • May 2008: 525–552
INTRODUCTION The traditionally circumscribed family Icacinaceae s.l.
includes about 55 genera and 400 species (Kårehed, 2001). Representatives are small shrubs to tall trees or lianas that typically grow in humid lowland forests throughout the trop- ics. A small number of species prefer subtropical forests and savannas to more temperate areas of Africa, Asia, Australia and South America, while others sporadically occur in salt marshes or in high-altitude regions between 2,000–3,000 m (Guymer, 1984; Howard, 1942; Sleumer, 1972; Villiers, 1973; de Roon & Mori, 2003; de Stefano, 2004). Malesia is the main centre of diversity comprising about half of the genera, followed by (sub)tropical America (20%–25% of the genera) and Africa (15%). Many genera are endemic to major phytogeographical areas, with the pantropical genus Citronella being one of the few exceptions.
Frequent mistakes in species determination, complex nomenclatural issues and undersampled collections make
Icacinaceae s.l. a poorly understood family of flowering plants (Howard, 1942; Sleumer, 1942a, b, 1972). Conse- quently, the taxonomic history shows much controversy (Table 1), and the family has been linked to various groups, including Olacaceae (De Candolle, 1824; Bentham, 1862) and the families Celastraceae and Aquifoliaceae (Miers, 1852; Engler, 1893). Subsequent authors followed the Celastrales connection (Rosidae, Cronquist 1981; Car- diopteris excluded), while Takhtajan (1997) recognised Icacinales as an order of its own in the Rosidae, including Aquifoliaceae, Icacinaceae (excluding Cardiopteris and Metteniusa), Phellinaceae and Sphenostemonaceae. Progress in molecular systematics has revealed that the Aquifoliaceae link is correct, but only for certain genera within Icacinaceae s.l. (Kårehed, 2001). A combined data set of ndhF, rbcL, atpB and 18S rDNA together with morphology showed that the traditional Icacinaceae are polyphyletic, and must be divided into four different families within asterids sensu APG II (2003): Icacinaceae s.str.
The wood anatomy of the polyphyletic Icacinaceae s.l., and their relationships within asterids
Frederic Lens1,2*, Jesper Kårehed3, Pieter Baas2, Steven Jansen4, David Rabaey1, Suzy Huysmans1, Thomas Hamann2 & Erik Smets1,2
1 Laboratory of Plant Systematics, Institute of Botany and Microbiology, Kasteelpark Arenberg 31, K.U. Leuven, 3001 Leuven, Belgium. *[email protected] (author for correspondence)
2 National Herbarium of the Netherlands—Leiden University Branch, P.O. Box 9514, 2300 RA Leiden, The Netherlands
3 The Bergius Foundation at the Royal Swedish Academy of Sciences and Department of Botany, Stockholm University, 106 91 Stockholm, Sweden
4 Jodrell Laboratory, Royal Botanic Gardens Kew, Richmond, Surrey TW9 3DS, U.K.
Wood samples from 53 species belonging to 41 genera of the Icacinaceae s.l. are investigated using light and scanning electron microscopy. The traditionally circumscribed Icacinaceae fall apart into four segregate families that are clearly nested within asterids, i.e., Icacinaceae s.str. (near or in Garryales), Cardiopteridaceae and Stemonuraceae (both Aquifoliales), and Pennantiaceae (Apiales). From a wood anatomical point of view, these families cannot easily be distinguished from each other. However, some features such as vessel distribution, perforation plate morphology, size and arrangement of vessel pits, fibre wall thickness, and the occurrence of cambial variants can be used to assign various species to one of the four families. The wood structure of the four segregate families is in general agreement with their suggested putative relatives, but the occurrence of lianas versus erect trees and shrubs is a confusing factor in getting clear phylogenetic signal from the wood structure. Maximum parsimony and Bayesian analyses using molecular data and combined anatomical-molecular data show that Icacinaceae s.str. are not monophyletic, and their closest relatives remain unclear. The combined analyses provide moderate support for a clade including Cassinopsis, the Apodytes-group, the Emmotum-group (all Icacinaceae s.str.), and the genus Oncotheca. This clade is situated at the base of lamiids and may be closely related to Garryales. The remaining lineage of Icacinaceae s.str., the Icacina-group represented by many climbing taxa exhibiting cambial variants, is strongly supported and might be sister to the rest of lamiids.
KEYWORDS: cambial variants, comparative wood anatomy, Garryales, Icacinaceae, LM, Oncotheca, SEM
MORPHOLOGY
526
TAXON 57 (2) • May 2008: 525–552Lens & al. • Wood anatomy of Icacinaceae
Table 1. Taxonomic overview of the genera studied according to their family classification sensu Miers (1852), Engler (1893), Bailey & Howard (1941a), Sleumer (1942ab) and Kårehed (2001).
Genus Miers (1852) Engler (1893) Bailey & Howard (1941b)
Sleumer (1942a, b) Kårehed (2001)
Apodytes Icacineae Icacineae Icacineae (I) Icacineae Icacinaceae s.str. (APO) Calatola – – Icacineae (I) Icacineae Icacinaceae s.str. (EMM?) Cantleya – – Icacineae (II) Icacineae Stemonuraceae +Cardioppteris+ – Cardiopterygoideae – Peripterygiaceae* Cardiopteridaceae Cassinopsis – Icacineae Icacineae (I) Icacineae Icacinaceae s.str. (CAS) +Chlamyydocaryya+ – Phytocreneae Phytocreneae (III) Phytocreneae Icacinaceae s.str. (ICA) Citronella – Icacineae Icacineae (I) Icacineae Cardiopteridaceae Codiocarpus – – – – Stemonuraceae Dendrobangia – – Icacineae (I) Icacineae Cardiopteridaceae Desmostachys Icacineae Icacineae Icacineae (III) Icacineae Icacinaceae s.str. (ICA) Discophora Sarcostigmateae – Icacineae (II) Icacineae Stemonuraceae Emmotum Emmoteae Icacineae Icacineae (I) Icacineae Icacinaceae s.str. (EMM) Gastrolepis – – Icacineae (II) Icacineae Stemonuraceae Gomphandra – – – Icacineae Stemonuraceae Gonocaryum Sarcostigmateae Icacineae Icacineae (II) Icacineae Cardiopteridaceae Hartleya – – – – Stemonuraceae +Icacina+ Icacineae Icacineae Icacineae (III) Icacineae Icacinaceae s.str. (ICA) +Iodes+ – Iodeae Iodeae (III) Iodeae Icacinaceae s.str. (ICA) Lasianthera – Icacineae Icacineae (II) Icacineae Stemonuraceae +Laviggeria+ – Icacineae Icacineae (III) Icacineae Icacinaceae s.str. (ICA) Leptaulus – Icacineae Icacineae (II) Icacineae Cardiopteridaceae +Mappppia+ Icacineae Icacineae Icacineae (III) Icacineae Icacinaceae s.str. (ICA) +Mappppianthus+ – Iodeae Iodeae (III) Iodeae Icacinaceae s.str. (ICA) Medusanthera – – Icacineae (II) Icacineae Stemonuraceae Merrilliodendron – – Icacineae (III) Icacineae Icacinaceae s.str. (ICA) Metteniusa – – – Icacineae Cardiopteridaceae? Oecopetalum – – Icacineae (I) Icacineae Icacinaceae s.str. (EMM) Ottoschulzia – – Icacineae (I) Icacineae Icacinaceae s.str. (EMM) Pennantia Sarcostigmateae Icacineae Icacineae (I) Icacineae Pennantiaceae +Phyytocrene+ – Phytocreneae Phytocreneae (III) Phytocreneae Icacinaceae s.str. (ICA) Platea – Icacineae Icacineae (I) Icacineae Icacinaceae s.str. (EMM?) +Polypyporandra+ – Iodeae Iodeae (III) Iodeae Icacinaceae s.str. (ICA) Poraqueiba Icacineae Icacineae Icacineae (I) Icacineae Icacinaceae s.str. (EMM?) Pseudobotrys – – – Icacineae Cardiopteridaceae? +Pyyrenacantha+ – Phytocreneae Phytocreneae (III) Phytocreneae Icacinaceae s.str. (ICA) +Rhapphiostyylis+ Icacineae Icacineae Icacineae (III) Icacineae Icacinaceae s.str. (APO) Rhyticaryum – Icacineae Icacineae (III) Icacineae Icacinaceae s.str. (ICA) +Sarcostiggma+ Sarcostigmateae Sarcostigmateae Sarcostigmateae (III) Sarcostigmateae Icacinaceae s.str. (ICA) Stemonurus Sarcostigmateae Icacineae Icacineae (II) Icacineae Stemonuraceae Whitmorea – – – – Stemonuraceae I, II and III refer to the groups of Bailey & Howard (1941b); APO, Apodytes group; CAS, Cassinopsis; EMM, Emmotum group; ICA = Icacina group. Names of climbing genera are marked with “+”. * Peripterygiaceae = Cardiopteridaceae.
527
(near or in Garryales, lamiids [euasterids I]), Cardiopteri- daceae and Stemonuraceae (both in Aquifoliales, campanulids [euasterids II]), and Pennantiaceae (in Apiales, campanulids) (Kårehed, 2001). However, it is possible that the present delimitation of Icacinaceae s.str., including the Apodytes group, the Emmotum group, the Icacina group and the genus Cassinopsis, does not constitute a monophyletic entity (Table 1). Also the position of Metteniusa and Pseudobotrys within Cardiopteridaceae remains open for discussion (Kårehed, 2001).
Anatomical characters of the stem have been considered to represent key features for the intrafamily classification. In order to define the four tribes of the subfamily Icacinoideae (i.e., Icacineae, Iodeae, Sarcostigmateae and Phytocreneae), Engler (1893) used habit, type of vessel perforation, wood development in young stems, and presence of interxylary (included) phloem in the wood (the same author also recognised two monogeneric sub- families: Cardiopteridoideae and Lophopyxioideae; the genus Lophopyxis is now included as a separate family in Malpighiales; Savolainen & al., 2000; APG II, 2003). Although Sleumer (1942a) followed the Icacinoideae classification, Bailey & Howard (1941a) pointed out that the four tribes proposed could not always be distinguished from each other based on the given characters. Therefore, Bailey & Howard (1941b) suggested an adjusted version of Engler’s classification of Icacinoideae based on the nodal anatomy and the perforation plate morphology, resulting in three groups: group I (trilacunar, exclusively scalariform perforations), group II (trilacunar, simple and scalariform perforations) and group III (unilacunar, exclusively simple perforations) (Table 1). However the same authors (1941b: 184) questioned the validity of their own classification by stating that “It should not be inferred from this, however, that all of the genera in one of these categories [i.e., their three groups] are necessarily more closely related geneti- cally to one another than to genera in the other categories.” This statement appears to be true based on molecular data (Kårehed, 2001).
Based on predominantly juvenile twigs of 50 genera and more than 150 species, the wood structure of Icacinaceae s.l. has been intensively studied by Bailey & Howard (1941b–d). The authors found an enormous variation in the secondary xylem: taxa with wood characters traditionally regarded as primitive according to Bailey & Tupper (1918), such as solitary vessels, scalariform perforations with many bars and long vessel elements, as well as taxa that possess a so-called derived set of wood features, such as a high frequency of vessel groupings, simple perforations and short vessel elements, are present. Moreover, much variation is present in the distribution of the axial parenchyma, the structure of the rays, and the presence of so-called cambial variants (Bailey & Howard, 1941b–d). Nevertheless, this series of studies must be in-
terpreted with caution because of the large number of juvenile wood samples included. It is generally known that the wood structure of juvenile twigs differs in many aspects from the wood of mature branches or trunks of the same plant. Not only quantitative characters, such as vessel diameter, vessel density, and ray width depend on this, also qualitative characters such as the initiation of specific cambial variants (amongst others successive cambia and interxylary phloem) only develop when a stem reaches a certain age (Obaton, 1960; Metcalfe & Chalk, 1983; Ursem & ter Welle, 1989; Carlquist, 2001). Next to the anatomical studies of Bailey & Howard, several other papers described the wood structure of a restricted number of Icacinaceae s.str. (Chalk & al., 1935; Metcalfe & Chalk, 1950; Détienne & al., 1982; Détienne & Jacquet, 1983; ter Welle & Détienne, 1994; Utteridge & al., 2005), Cardiopteridaceae (Metcalfe & Chalk, 1950; Normand, 1950; Détienne & al., 1982; Détienne & Jacquet, 1983; Bamber & ter Welle, 1994; ter Welle & Détienne, 1994), Stemonuraceae (Howard, 1943; Metcalfe & Chalk, 1950; Détienne & Jacquet, 1983; ter Welle & Détienne, 1994) and Pennantiaceae (Metcalfe & Chalk, 1950; Meylan & Butterfield, 1978; Patel & Bowles, 1978). Additional descriptions and micrographs of 21 genera belonging to all four families can be found on the InsideWood website (http://insidewood.lib.ncsu.edu/search). Wood anatomical papers that dealt with cambial variants of a small number of climbing Icacinaceae s.l. include Engler (1893), Sleumer (1942a), Obaton (1960), Ursem & ter Welle (1989), Bamber & ter Welle (1994), and Utteridge & al. (2005).
Major objectives of this study are: (1) providing a wood anatomical overview of the traditionally circumscribed Icacinaceae s.l. using original observations and literature data, (2) searching for wood anatomical synapomorphies that could define the segregate families Icacinaceae s.str., Cardiopteridaceae, Stemonuraceae and Pennantiaceae, (3) relating the wood anatomical diversity of the segregate families to their asterid relatives, and (4) evaluating the phylogenetic significance in the wood structure of the Icacinaceae segregates by carrying out phylogenetic analyses based on existing molecular and combined molecular-anatomical data at the asterid level.
MATERIAL AND METHODS Wood anatomical descriptions and microtech-
niques. — In total, 61 wood specimens of Icacinaceae s.l., including 53 species and 41 genera, were investigated using light microscopy (LM) and scanning electron microscopy (SEM) (Appendix 1). Despite the undersampling of the traditional Icacinaceae in most botanical collections, the number of genera studied represents a good coverage
of the segregate families at the genus level: Icacinaceae s.str. (23/34, including all lineages sensu Kårehed, 2001), Cardiopteridaceae (7/7), Stemonuraceae (10/12) and Pen- nantiaceae (1/1). Most samples were derived from mature sapwood, except for the juvenile twigs of Cardiopteris moluccana, Chlamydocarya thomsonia, Gastrolepis austro-caledonica, Icacina claessensi, Lavigeria macrocarpa (BR and Kw specimen in Appendix 1), Leptaulus grandifolius, Mappianthus iodoides, Pyrenacantha lebrunii, and Pyrenacantha kirkii. In order to increase our sam- pling, we were able to study a selection of the original slides of Bailey & Howard (1941a–d), including mostly juvenile stems of the genera Cassinopsis, Chlamydocarya, Grisol- lea, Hosiea, Icacina, Iodes, Lavigeria, Mappia, Mique- lia, Natsiatum, Notha podytes, Oecopetalum, Phyto crene, Pittosporopsis, Pleurisanthes, Pyrenacantha, Rhaphiosty- lis and Sarcostigma. Alcohol preserved material and fresh stems of lianas were requested from botanical gardens in order to make better sections, because most climbing taxa exhibit cambial variants including nonlignified tis- sues such as parenchyma and phloem (see Appendix 1). The specimen observed of Pyrenacantha malvifolia, a fast growing liana from the greenhouses of the Botanic Gar- dens in Kew, was omitted from the descriptions and Table 2 because of the huge amount of unlignified parenchyma and poorly developed fibres. This anatomical deviation is typical of fast growing plants in greenhouses, and should be treated with caution in comparative studies.
The methodology of wood sectioning and the subsequent steps are described in Lens & al. (2005b). The wood anatomical terminology follows the “IAWA list of microscopic features for hardwood identification” (IAWA Committee, 1989). We define tracheids as long, imper- forate cells with more than one row of distinctly bordered pits in tangential and radial walls (usually between 8–10 μm in horizontal diameter), or with only one row of very large conspicuously bordered pits (often more than 10 μm in horizontal diameter). When distinctly bordered vessel-ray pits are mentioned, we mean that the pit pairs are half-bordered, because the pit on the vessel side is bordered, but predominantly simple on the parenchyma cell side.
Phylogenetic analysis. — In total, 26 wood anatomical characters (Appendix 2 in Taxon online issue) are combined with the existing molecular asterid matrix of Kårehed (2001) based on ndhF, rbcL, atpB and 18S rDNA, comprising in total 6,950 molecular characters, of which 2,251 are parsimony informative. A few adjust- ments have been carried out on the original Kårehed matrix: (1) all non-asterid taxa are omitted from the analysis because of the clear asterid origin of Icacinaceae s.l., (2) the original morphological data for Icacinaceae s.l. have been excluded, (3) only one end taxon per genus has been retained (except for Pennantia), excluding the taxa Cassi-
nopsis ilicifolia, Citronella moorei, Leptaulus daphnoides, and Pyrenacantha grandifolia, (4) complete wood and sequence data of Eucommia (GenBank accession numbers AJ235469, AJ429113, L01917, L54066) and Oncotheca (accession numbers AJ429114, AF206976, AJ131950, AJ235549) have been included in order to investigate their relationship with Icacinaceae s.str. and Garryales, and (5) some sequences that were lacking in the original matrix have been added from GenBank, such as ndhF sequences of Gonocaryum (AJ400889) and Medusanthera (unpub. data), 18S sequences of Gonocaryum (AF206919), rbcL of Apodytes (AJ428895), Cassinopsis (AJ428896), Gono- caryum (AJ235779), Pennantia corymbosa (AJ494842), P. cunninghamii (AJ494843) and Pyrenacantha (AJ235791), and atpB sequences of Gonocaryum (AJ400883), Pennan- tia corymbosa (AJ494840), P. cunninghamii (AJ494841) and Pyrenacantha (AJ235575).
In total, 109 asterid taxa are included in the data matrix, of which the seven Cornales taxa are designated as outgroup. The genera Anagallis, Boopis, Borago, Calli- triche, Campanula, Codonopsis, Dipsacus, Hydrophyl- lum, Lamium, Menyanthes, Nymphoides, Panax and Pro- boscidea are herbaceous, and are coded as unknown for all wood characters. Likewise, no data could be found for the woody genus Phyla, which is coded as unknown. Information from the other 95 genera was gathered using observations in Ericales (Lens & al., 2005a, b, 2007a, b) and Icacinaceae s.l. (this paper), while other groups were coded based on information from the InsideWood data- base (http://insidewood.lib.ncsu.edu/search/) and from ex- tensive literature data (Howard, 1943; Metcalfe & Chalk, 1950; Carlquist, 1957, 1969a, b, 1978, 1981a, b, 1982, 1983, 1984, 1985, 1986, 1991, 1992, 1997; Obaton, 1960; Baas, 1973, 1975; Carpenter & Dickison, 1976; Giebel & Dicki- son, 1976; Stern, 1978; Styer & Stern, 1978; Mennega, 1980; Dickison, 1982; Carlquist & al., 1984; Dickison & Phend, 1985; Carlquist & Hoekman, 1986; Gornall & al., 1988; Gregory, 1988a–c; Ogata, 1988; Baas & al., 1988; Carlquist & Zona, 1988; Liang & Baas, 1990; Schweingru- ber, 1990; Carlquist & Hanson, 1991; Gasson & Dobbins, 1991; Ogata, 1991; Oskolski, 1996; Oskolski & al., 1997; Noshiro & Baas, 1998).
We have compiled the wood matrix following the strategy of Lens & al. (2007b): (1) the selection of characters is based on their variation within asterids and their stabil- ity at generic level (Appendix 2 in Taxon online issue), which is similar to most characters proposed by Herend- een & Miller (2000); (2) highly variable characters were excluded from the analysis (e.g., distinctness of growth rings, horizontal diameter of intervessel pits, presence of two distinct ray sizes, number of cells in multiseriate rays at their widest portions, and incidence of sheath cells) or coded using major character states only (axial parenchyma distribution); (3) only one of two dependent characters was
taken into account, such as vessel diameter/density and vessel element/fibre length; (4) neotenous or paedomor- phic characters (chars. 5, 10, 16–18), defined as juvenile characters of the primary xylem that have been protracted into the secondary xylem (Carlquist, 1962) and typical of taxa with secondarily derived wood, were not taken into account and were coded with a “?” (Appendix 2 in Taxon online issue); (5) ecologically dependent features (chars. 1, 3–4, 6, 8–10, 14) and habit related characters (chars. 3–4, 9) were not excluded from the matrix due to their possible phylogenetic value; (6) the continuously varying characters (chars. 4, 9–10) were coded following the gap weighting method of Thiele (1993) resulting in ordered, multistate characters with ten possible states and weight of 0.1 for each of the three characters; and (7) two additional multistate wood features are ordered (chars. 3, 5).
Maximum parsimony (MP) analyses were conducted with PAUP*4.0b10 (Swofford, 2002), employing a heuristic search with 2,500 random addition replicates, TBR branch swapping (one tree held at each step, MULTREES off). The shortest tree from each replicate was saved, and these trees were subsequently submitted to a second round of TBR swapping with the MULTREES option on. Bootstrap analyses were carried out using 10,000 bootstrap replicates, a random addition sequence with five replicates, and TBR swapping (one tree held at each step, MULTREES off). In order to assess the phylogenetic value of wood characters, both the adjusted molecular matrix of Kårehed (2001) and the combined molecular/anatomical matrix were analysed following the same strategy.
A Bayesian inference of phylogeny was conducted with the software programme MrBayes version 3.1.2 that calculates posterior probabilities using the Markov chain Monte Carlo (MCMC) algorithm (Huelsenbeck & Ron- quist, 2001). The wood anatomy data and the indel data of the ndhF matrix were analysed under the standard discrete (morphology) model (Posada & Crandall, 1998). The Mor- phoCode characters were coded as unordered, because MrBayes allows only six different character states for one ordered character (the shift from ordered to unordered MorphoCode characters does not affect the topology in MP analyses). Models for nucleotide substitution were evaluated using MrAIC, version 1.4.2 (Nylander, 2004). The ndhF matrix was consequently analysed under the general time reversible model (GTR) with a gamma distribution of substitution rates. The other molecular markers were analysed under the same model but in addition with a proportion of invariant sites. When running the analyses, the dataset was partitioned and the partitions unlinked, so each had its own set of parameters. The Markov chain was run for 5,000,000 generations and every 100th tree was sampled. Three additional “heated” chains were used for each run (Metropolis-coupled Markov chain Monte Carlo; Huelsenbeck & Ronquist, 2001). At least two separate runs
for each dataset were performed to evaluate if the chain had become stationary. Of the resulting 50,000 trees, the first 5,000 (burn-in) were discarded when calculating the posterior probabilities.
RESULTS Wood descriptions. — The material studied is de-
scribed according to the recent family classification of Kårehed (2001). For each genus examined, the numera- tor represents the number of species studied and the de- nominator includes the total number of species. Numbers without parentheses are ranges of means, while numbers between parentheses represent minimum or maximum values. Descriptions of continuous characters are based on mature wood samples. An asterisk after the genus names refers to genera with climbing species. A summary of selected wood features is shown in Table 2.
Icacinaceae s.str. (lamiids, near or in Garryales) (Apo- dytes 2/ca. 17, Calatola 2/8, Cassinopsis 2/4, Chlamydo- carya* 1/7, Desmostachys 1/7, Emmotum 1/7, Icacina* 2/5, Iodes* 1/24, Lavigeria* 1/1, Mappia* 1/7, Mappianthus* 1/1, Merrilliodendron 1/1, Oecopetalum 1/2, Ottoschulzia 1/3, Phytocrene* 1/19, Platea 2/8, Polyporandra* 1/1, Poraqueiba 1/3, Pyrenacantha* 3/24, Rhaphiostylis* 1/10, Rhyticaryum 1/18, Sarcostigma* 1/6; Figs. 1–4): Growth ring boundaries usually indistinct to completely absent (Fig. 1A–D), indistinct in Cassinopsis and Emmotum, and distinct in the young stems of Icacina claessensi and Mappianthus iodoides. Diffuse-porous. Vessels (0–)3–85 (–103)/mm2, mostly solitary in all species (Figs. 1A–B), also a small percentage (usually 5%–25%) of radial multiples of 2–3(–4) in species of Desmostachys, Merrilliodendron, Ottoschulzia, Phytocrene, Platea, Rhaphiostylis and Rhyti- caryum, additional tangential multiples (often between 10%–30%) of 2–3(–4) present in Chlamydocarya, Des- mostachys (Fig. 1C), Icacina, Iodes, Lavigeria, Mappia, Merrilliodendron, Phytocrene, Polyporandra (Fig. 1D), Rhaphiostylis and Sarcostigma; vessel outline rounded to elliptical in the climbing species, but mostly slightly angular in the nonclimbing ones; perforation plates predominantly or entirely simple in most genera studied, but exclusively scalariform with (6–)8–80(–117) bars in Apodytes, Calatola, Cassinopsis (Fig. 1E), Emmotum, Oecopetalum, Ottoschulzia, Platea and Poraqueiba, exceptionally with reticulate portions in Apodytes. Intervessel pits opposite to alternate in species with predominantly simple perforations, pits 5–11 μm in horizontal diameter, intervessel pits scalariform to opposite in species with scalariform perforations (except for Rhaphiostylis), pits 5–40 μm in horizontal diameter, pit apertures coalescent in Icacina, Iodes, Lavigeria and Sarcostigma, nonvestured. Vessel-ray pits similar to intervessel pits in size and shape; vessel-
530
Table 2. Overview of selected wood anatomical characters within Icacinaceae s.l.
Species S ol
2 )
+Chlamyydocaryya thomsoniana+* ICA + – ± > 95 1–3–13 (O)–A 7–8 – 25–70–150 44–55–68 Desmostachys vogelii ICA + ± + 100 – O–A 5 – 40–100–160 5–8–12 +Icacina claessensi+* ICA + – ± 100 – O–A 6–8 – 40–120–220 11–14–17 +Icacina mannii+ ICA + – ± 100 – (O)–A 6–9 – 35–155–250 6–9–12 +Iodes affricana+ ICA + – ± 100 – (O)–A 8–10 – 50–215–400 6–12–20 +Laviggeria macrocarppa+3 ICA + – ± > 95 1–9–21 O–A 7–8 – 15–110–180 12–18–23 +Mappppia cordata+1 ICA + – + 100 – (O)–A 6–9 – 15–150–340 6–11–14 +Mappppia cordata+2 ICA + – ± 100 – A 8–10 – 120–200–250 6–10–13 +Mappppianthus iodoides+* ICA + – – 100 – O–A 4–5 – 35–70–80 20–35–47 Merrilliodendron megacarpum ICA + ± + 100 – (O)–A 5–6 – 25–70–105 6–9–17 +Phyytocrene macropphyylla+ ICA + – ± 100 – O– (A) 7–9 – 25–310–650 4–7–9 +Phyytocrene spp.+ ICA + ± ± 100 – O–A 6–8 – 25–180–475 9–16–22 +Polypyporandra scandens+ ICA + – ± 100 3–6–11 O–A 5–6 – 55–120–210 5–12–19 +Pyyrenacantha lebrunii+* ICA + – – 100 – O–A 5–6 – 15–50–100 56–65–84 +Pyyrenacantha kirkii+* ICA + – ± 100 – A 6–8 – 20–42–60 120–140–160 Rhyticaryum longifolium ICA + ± – 100 – O–A 7–10 – 60–80–110 0–3–6 +Sarcostiggma kleinii+ ICA + – ± 100 – O–A 7–10 – 35–120–250 7–10–16 Cassinopsis capensis CAS + – – 0 13–25–31 O–A 6–8 – 20–34–50 67–85–103 Cassinopsis sp. CAS + – – 0 39–70–117 S–O 5–12 + 75–90–125 21–25–31 Cassinopsis tinifolia CAS + – – 0 32–40–49 (S)–O 5–7 ± 32–55–70 31–40–59 Calatola columbiana EMM + – – 0 21–40–57 O 5–6 – 50–80–115 8–13–17 Calatola venezuelana EMM + – – 0 23–40–55 O– (A) 5–6 – 40–60–75 20–25–30 Emmotum fagifolium EMM + – – 0 6–9–11 O 5–7 – 80–115–150 11–13–16 Oecopetalum mexicanum EMM + – – 0 17–30–43 S–O 5–17 – 20–50–75 25–40–55 Ottoschulzia pallida EMM + ± – 0 6–8–11 O– (A) 5–6 – 25–45–70 25–33–38 Platea excelsa EMM + ± – 0 32–55–99 S– (O) 5–6 – 70–85–115 20–25–32 Platea hainanensis EMM + + – 0 34–80–110 S–O 8–40 – 50–75–100 22–29–32 Poraqueiba guianense EMM + – – 0 14–18–22 O 5–7 ± 80–120–155 6–11–14 Apodytes dimidiata APO + – – 0 23–30–35 O 5–7 – 55–75–95 12–17–22 Apodytes javanica APO + – – 0 20–30–37 O 6–8 – 60–75–95 17–21–27 +Rhapphiostyylis f ferrugginea+1 APO + – ± 100 0 O–A 7–11 – 50–180–360 6–10–12 +Rhapphiostyylis fferrugginea+2 APO + ± ± 100 0 O–A 9–11 – 80–245–370 7–9–10 +Cardioppteris moluccana+* CAR + – ± 100 0 O– (A) 9–14 – 50–260–475 4–6–9
531
Ve ss
el e
le m
am bi
al v
ar ia
nt s
y y 225–370–600 400–750–1,200 – + + + + 2–3 200–335–875 r r – – – fx, sc 500–730–1,000 1,100–1,630–2,200 – ± ± + + 2–6 1,000–3,350–7,500 r – – – r – 250–540–1,100 750–1,020–1,500 – + + ± + 4–6 2,300–3,780–7,100 – – – – – – 250–860–1,200 900–1,360–1,900 – ± + ± + 5–32 (40) > 5,000 r, ap – – – r sc
f 300–495–725 600–960–1,200 – ± + – + 2–6 (20) 200–1,300– >8,000 r r – – r sc g p 400–690–950 700–1,070–1,600 – ± ± – + 6–10 (20) > 9,000 – – – – – sc
pp 500–880–1,150 1,100–1,540–2,300 – + ± – + 2–10 > 8,000 r, ap – – – – sc pp 300–425–550 700–930–1,100 – + + ± + 2–3 (6–9) 200–830–2,100 r – – – – sc pp 550–790–850 650–1,130–1,450 – + + + ± 2–7 > 6,000 – – – – – –
200–450–700 700–970–1,300 – + + – + 2–4 300–800–1,650 r, ap r – r r – y p y 200–290–450 350–650–1,100 – + ± – + 2–42 > 1,2000 – r – – r fx, sc y p 70–175–300 250–610–1,000 – ± + – + 2–19 150–1,120–3,700 r, ap r – – – fx, sc yp 375–735–950 760–1,070–1,450 – – + – + (2) 12–27 > 7,000 r, ap r, ap – – – –
y 150–240–325 225–490–950 – ± ± + ± 2 150–225–400 – – – – r fx y 150–250–350 400–550–750 – + – + – – – – – – – – fx
350–500–650 800–1,500–2,300 – ± + – ± 2–8 800–1,600–3,200 – – – r – – g 100–325–500 300–870–1,200 – ± + – + 2 200–365–700 r, ap – – r r ip
600–1,130–1,450 1,200–1,700–2,000 + + – + – (2–4) 5–10 500–1,230–1,800 r – – – – – 1,100–2,220–2,900 2,500–3,120–3,800 ± + – ± – 2–5 900–2,090–5,900 – – – – – – 900–1,330–1,700 1,600–1,900–2,200 – + – ± – 2–3 (4) 200–770–1,800 – – – – – –
1,400–1,790–2,800 1,800–2,500–3,000 – + ± ± – 3–5 > 5,000 r – – r – – 1,000–1,490–1,950 2,000–2,410–2,700 – + ± + – 3–4 900–1,730–2,300 r – – – – – 900–1,390–1,950 1,800–2,300–2,850 – + ± ± – 2–4 (16) 50 –750 – > 9,000 r – – – – – 900–1,445–2,400 1,450–2,050–2,900 ± ± ± ± – 2–5 800–1,980–5,900 r – – – – – 650–840–1,150 1,000–1,680–2,000 – + – + – 10–27 2,000–5,500–1,0000 r – – – – –
1,000–1,495–2,100 1,800–2,170–2,600 – + ± ± – 2–5 500–870–1,550 – – – – – – 1,300–2,000–2,700 2,300–2,860–3,300 – + – + – 3–5 700–1,460–2,600 r – – – – – 1,200–1,620–2,100 1,900–2,400–2,900 – + ± ± – 2 (9–21) 50–2,450 – > 7,000 r – – – – – 1,200–1,760–2,200 2,250–2,880–3,300 – + – ± – 2–3 300–770–1,200 – – – r r – 1,200–1,800–2,400 2,300–2,910–4,700 – + – ± – 2–4 350–1,110–3,300 – – – – – –
p y f g 500–680–900 800–1,170–1,700 – + – – + (2–4) 4–8 350–1,110–4,250 r, ap – – – – sc p y f g 500–700–950 750–1,150–1,500 + + – ± + (2–3) 4–6 > 1,0000 ap – – – – sc
p 300–470–950 1,100–1,600–2,200 – – + + ± 2 200–365–650 – r – – – pgt
532
Table 2. Overview of selected wood anatomical characters within Icacinaceae s.l.
Species S ol
2 )
Citronella silvatica CAR + – – 0 37–50–65 O 5–8 – 60–100–140 16–25–32 Dendrobangia boliviana CAR + – – 0 25–30–40 O 5–6 – 115–135–155 3–5–7 Gonocaryum crassifolium CAR + – – >95 0 O–A 5–7 – 50–70–110 18–25–31 Leptaulus daphnoides CAR + – – 5–10 1–2–6 O 5–7 – 45–55–67 48–55–73 Leptaulus grandifolius* CAR + – ± >60 1–2–5 O– (A) 5–6 – 20–25–32 102–130–156 Metteniusa cf. edulis CAR + – – 0 24–33–50 O– (A) 5–6 + 30–70–90 19–24–29 Pseudobotrys dorae CAR + – – 0 17–35–51 (S)–O 5–10 – 30–40–55 32–45–63 Cantleya corniculata STE + – – 100 0 O 8–9 + 75–105–125 5–7–10 Codiocarpus merrittii STE + + – >95 1–12–22 O– (A) 5–6 – 45–65–85 31–33–39 Discophora guianensis STE + + – 80 3–10–21 O–A 5–8 – 50–75–105 20–27–36 Gastrolepis austro-caledonica* STE + ± – 50 1–5–15 (S)–O 6–9 + 40–50–60 80–95–103 Gomphandra luzoninesis STE + – – 75 1–18–64 O 5–7 – 60–100–120 5–7–10 Hartleya inopinata STE + ± – 50 1–5–30 S–O 6–12 + 55–75–95 11–18–21 Lasianthera africana STE + – – >90 1–3–5 O–A 5–6 + 30–45–55 34–45–50 Lasianthera apicalis STE + + – >70 1–7–44 S–O 6–13 – 80–135–175 9–12–15 Medusanthera laxiflora STE + ± ± >95 1–10–37 S–O– (A) 5–6 + 65–90–115 13–19–25 Stemonurus celebicus STE + ± ± >90 1–5–38 S–O 8–34 – 95–130–175 11–13–15 Stemonurus mallacensis STE + ± + 75 1–10–34 S–O– (A) 8–20 + 75–110–135 11–15–22 Stemonurus umbellatus STE + + + 80 8–16–25 S–O 10–35 – 80–115–190 11–15–18 Whitmorea grandiflora STE + ± – >90 1–5–21 S–O 8–13 – 65–80–100 4–7–10 Pennantia corymbosa PEN + ? ? 0 18–28–45 O 3–4 – 30–40–55 69–85–112 Pennantia cunninghamii PEN + – – 0 33–40–56 O 3–4 – 50–70–95 48–54–60 Species are arranged alphabetically according to the family classification of Kårehed (2001). Numbers between hyphens are mean values, flanked by minimum and maximum values; numbers/symbols between parentheses represent exceptional values/condi- tions; measurements for horizontal diameter of intervessel pits and multiseriate rays width are ranges. For specimens of the same taxon, superscript numbers after the species name refer to the order of the specimens in the species list (see Appendix 1).
Table 2. Continued.
ray pits sometimes scalariform with reduced borders to even simple in species of Cassinopsis and Poraqueiba (Fig. 1F); vessel-ray pitting occasionally unilaterally compound in Cassinopsis tinifolia, Merrilliodendron and Phytocrene. Wall sculpturing present as short fine ridges in the inner vessel wall of Desmostachys and Iodes (Fig. 2A). Tyloses occasionally present in Icacina and Mappia. Tangential diameter of vessels (20–)35–310(–650) μm, ves-
sel elements (70–)175–2,200(–2,900) μm long. Tracheids present in Chlamydocarya, Icacina, Iodes, Lavigeria, Polyporandra, Phytocrene, Rhaphiostylis (Fig. 2B) and Sarcostigma, (250–)600–1,000(–1,500) μm long. Nonsep- tate fibres with large bordered pits (Fig. 2C) concentrated in tangential and radial walls, co-occurring with septate fibres having simple pits in Sarcostigma (Fig. 2D), fibres usually very thin- or thin- to thick-walled (thick-walled in
533
Poraqueiba), (350–)610–3,120(–3,800) μm long, pit borders (3–)5–8(–10) μm in diameter. Axial parenchyma in nonclimbing genera often a combination of diffuse-in-aggregates and scanty paratracheal parenchyma with a tendency to form short narrow bands in Calatola, Desmostachys (also wider bands, Fig. 1C), Emmotum, Merrilliodendron, Oecopetalum, Platea, and Poraqueiba (Fig. 1B), banded parenchyma 3–5 cells wide dominant in Rhyticaryum,
vasicentric parenchyma and conspicuous band formation of (1–)2–10(–20) cells wide dominant in the climbing genera (Figs. 1D, 3E, bands absent in Raphiostylis), strands including 4–16 cells. Uniseriate rays scarce to abundant, (0–)1–29(–33) rays/mm, (30–)85–925(–2,500) μm long, consisting of square to upright cells. Multiseriate rays in nonclimbing genera generally 2–5-seriate (Fig. 2E–F, few multiseriate rays much wider in Emmotum (up to 16-seri-
Ve ss
el e
le m
am bi
al v
ar ia
nt s
850–1,390–1,950 1,900–2,520–3,100 – + – + – 4–7 > 5,000 r – – r – – 1,100–2,230–2,800 3,300–4,240–4,900 – + ± + – (2–3) 5–8 > 6,000 r, ap – – r – –
500–960–1,425 2,600–2,970–3,450 – + ± ± – (3–4) 5–9 650 – > 6,000 r – – r – – 500–795–1,075 1,700–2,190–2,900 – + – + – 2–4 300–620–950 – – – r – –
650–1,095–1,600 1,300–1,550–1,800 ± ± – + – 2–3 700–2,450–5,200 – – – – – – 1,200–1,800–2,200 2,500–2,870–3,300 – + + + – 5–11 1,300–2,500–3,400 r – – – – – 1,025–1,310–1,700 2,150–2,770–3,150 – + – ± – (6) 11–19 > 6,000 r – – r – – 900–1,430–1,800 1,750–2,250–2,550 ± – – + – 2–3 400–1,200–3,875 – – – r – – 400–785–1,050 2,250–2,790–3,200 – + – + – 2–5 300–835–1,600 – – – r – –
100–1,030–1,850 3,000–3,650–4,400 – + + + – (2) 3–7 1,000–2,810–4,400 r – r r – – 275–540–900 850–1,160–1,700 ± – – + – 2–4 350–860–1,500 r – – – r –
1,000–1,410–1,850 2,700–3,370–4,150 – – + ± – (6) 10–17 4,000–8,000–9,000 – – r r – – 600–1,420–1,950 1,450–3,020–4,500 – + – + – 2–5 325–1,090–2,700 – – r r – – 400–865–1,100 1,700–2,110–2,600 – + – + – 2–6 550–1,150–2,150 – – r – – –
950–1,510–2,000 2,700–3,330–4,000 ± – – + – 2–6 500–1,830–4,100 r – – – – – 300–600–800 1,100–2,450–3,300 – ± + + – 9–11 2,500 – > 6,000 r – – r r –
1,000–1,400–1,850 2,500–3,470–4,500 – ± – + – 2–6 700–1,930–3,800 r – – r r – 400–900–1,500 1,500–2,220–3,000 – ± – + – (2–4) 6–10 300–1,090–2,100 r – – r r –
600–1,130–1,600 1,700–2,460–3,100 – ± – + – (2–4) 5–9 1,200–2,080–3,100 r, f – – – – – 450–665–1,000 1,600–2,000–3,400 ± – – + – (2–4) 6–9 300–1,100–1,700 r – – r r –
700–1,080–1,400 1,300–1,790–2,300 – + – + – (2–3) 5–6 900–1,390–2,000 – – – – – – 900–1,500–2,250 1,700–2,270–2,900 – + – + – 4–8 550–1,020–1,750 – – – – – –
Names of climbing species are marked with “+”, juvenile wood samples with asterisks. Three-letter acronyms after genera refer to acronyms used in Table 1. Intervessel pit arrangement: A, alternate; O, opposite; S, scalariform. Type of crystals: ap, in axial parenchyma; f, in fibres; r, in rays. Type of cambial variants: fx, furrowed xylem; ip, included phloem; pgt, parenchymatous ground tissue; sc, successive cambia. + = present, ± = scarcely present, – = absent.
534
ate) and Poraqueiba (up to 21-seriate; Fig. 2G), and all multiseriate rays wider in Ottoschulzia (10–27-seriate); multiseriate rays in climbing genera extremely variable, but absent in the juvenile stem of Pyrenacantha kirkii, ranging from 2-seriate in Sarcostigma (Fig. 2D) up to 42-seriate in
Phytocrene macrophylla, rays partly or entirely unlignified in Icacina (Fig. 3B), Iodes, Lavigeria (Fig. 3C), Phytocrene (Fig. 3D), Polyporandra (Fig. 1D), Pyrenacantha, Mappia, Mappianthus and Rhaphiostylis, multiseriate ray height (50–)365–5,500( – > 12,000) μm, density (0–)1–11(–15)
Fig. 1. LM micrographs (A–D, F) and SEM micrograph (E) of the wood structure in Icacinaceae s.str. A, Apodytes dimidiata, transverse section (TS), solitary vessels and mainly diffuse-in-aggregates axial parenchyma; B, Poraqueiba guianense, TS, solitary vessels and mainly diffuse-in-aggregates axial parenchyma with slight tendency to form short bands; C, Desmostachys vogelii, TS, vessels solitary and in short tangential multiples, axial parenchyma mainly paratracheal; D, Polyporandra scandens, TS, wide axial parenchyma bands and wide multiseriate rays; E, Cassinopsis sp., radial longitudinal section (RLS), scalariform perforation plate; F, Poraqueiba guianense, RLS, vessel-ray pitting opposite to scalariform with reduced borders.
535
rays/mm, rays consisting of homocellular (exclusively procumbent) body ray cells in Apodytes, Emmotum, Poraqueiba, Platea, and heterocellular (a mixture of procumbent, square and/or upright) body ray cells in the other genera, 1–4( > 10) rows of upright to square marginal
ray cells; indistinct sheath cells sometimes present in Ca- latola, Cassinopsis (Fig. 2F), Desmostachys, Mappian- thus, Platea and Rhyticaryum; rays almost never fused. Dark amorphous contents sometimes present in Apodytes. Solitary prismatic crystals in body and marginal ray cells
Fig. 2. Tangential longitudinal surfaces (TLS, SEM, A–C) and sections (LM, D–G) of Icacinaceae s.str. A, Iodes africana, short oblique wall thickenings in the inside wall of a vessel element; B, Rhaphiostylis ferruginea, tracheid with one row of conspicuously bordered pits; C, Cassinopsis sp., fibre with distinctly bordered pits; D, Sarcostigma kleinnii, uniseriate rays and septate libriform fibres (two arrows at right) that co-occur with fibre-tracheids having distinctly bordered pits (arrow at left); E, Apodytes javanica, 2–3-seriate rays with a variable number of rows of marginal ray cells; F, Cassinopsis capensis, multiseriate rays showing indistinct sheath cells (arrows); G, Poraqueiba guianense, large multiseriate rays co- occurring with uniseriate ones.
536
(usually nonchambered, but partly chambered in Mappia and Sarcostigma, absent in Apodytes, Iodes, Rhyticaryum, Pyrenacantha) and in axial parenchyma cells of species of Icacina, Mappia (chambered), Merrilliodendron, Phy- tocrene, Polyporandra, Rhaphiostylis and Sarcostigma, few small styloid-like crystals in rays of Apodytes, Mer-
rilliodendron, Rhyticaryum and Sarcostigma, druses in nonlignified rays of Iodes, Phytocrene and Polyporandra, and in axial parenchyma of Polyporandra, few clustered crystals in nonchambered body and marginal ray cells of Apodytes, Desmostachys, Icacina, Iodes, Merrillioden- dron, Phytocrene, Pyrenacantha and Sarcostigma, crystal
Fig. 3. Transverse sections (TS, LM) showing cambial variants in Icacinaceae s.str. A, Sarcostigma kleinnii, interxylary phloem (arrow); B, Icacina mannii, successive cambia; C, Lavigeria macrocarpa, successive cambia with thick-walled sclereids in the conjunctive tissue (arrows); D, Phytocrene macrophylla, successive cambia with elongated phloem zones (arrow); E, Mappia cordata (Tw specimen), detail of phloem region in between two regions of wood produced by successive cambia; F, Icacina mannii, detail of secretory canals (arrows) in the conjunctive tissue.
537
sand absent. Cambial variants observed as three types, and more than one type may occur within a single species: (1) interxylary phloem forming islands of phloem within the wood cylinder (Sarcostigma, Fig. 3A); (2) successive cambia (Chlamydocarya [only in furrows, Fig. 4B], Iodes, Icacina [Fig. 3B], Lavigeria [Fig. 3C], Mappia [Fig. 3E],
Phytocrene [Fig. 3D], Pyrenacantha and Rhaphiostylis) with thick-walled sclereids in the conjunctive tissue, pro- liferation of parenchyma is obvious in species with successive cambia, especially in Iodes with abnormal position of the primary xylem; and (3) furrowed xylem with elongated zones of phloem in the furrows (Chlamydocarya
Fig. 4. Transverse sections (A–D) and tangential section (E) of cambial variants. A, Pyrenacantha lebrunii, entire juvenile stem with furrowed xylem, showing three well-developed (white arrows) and two initial (black arrows) phloem zones in the stem; B, Chlamydocarya thomsonia, successive cambia (arrows) constricted to the phloem zone; C, Phytocrene sp., elongated phloem zones separated by unlignified rays; D, Phytocrene sp., detail of phloem showing sieve tubes, companion cells, fibres and axial parenchyma; E, Phytocrene sp., sieve tubes with scalariform sieve plates.
538
[Fig. 4B], Phytocrene [Fig. 4D], Pyrenacantha [Fig. 4A]), in Phytocrene and Pyrenacantha phloem zones include mainly phloem fibres and tangential multiples of sieve tubes, together with few companion and phloem parenchyma cells (Fig. 4C–D); secretory canals in nonlignified conjunctive parenchyma and rays of Icacina mannii (Fig. 3F); phloem with typically scalariform sieve plates (Fig. 4E) and many druses.
Cardiopteridaceae (campanulids, Aquifoliales) (Cardiopteris* 1/3, Citronella 1/10, Dendrobangia 1/1, Gonocaryum 1/ca. 20, Leptaulus 2/6, Metteniusa 1/3, Pseudobotrys 1/2; Fig. 5): Growth ring boundaries often indistinct to absent, more obvious in Leptaulus, Metteniusa and Pseudobotrys. Diffuse-porous. Vessels (3–)5–55(–73)/ mm2, predominantly solitary (Fig. 5B) with occasionally short tangential multiples of 2–3 in Cardiopteris (Fig. 5A) and Leptaulus grandifolius; vessel outline (slightly) angular, but oval to rounded in Cassinopsis; perforation plates exclusively or predominantly scalariform with (17–)30–50(–65) bars in Citronella, Dendrobangia, Met- teniusa and Pseudobotrys, exclusively or predominantly simple in Cardiopteris and Gonocaryum, and simple/ scalariform with few bars in Leptaulus, few reticulate perforations in Gonocaryum (Fig. 5D), reticulate portions in Dendrobangia. Intervessel pits opposite in Citronella and Dendrobangia, opposite to slightly scalariform in Pseudobotrys, opposite to slightly alternate in species of Cardiopteris, Leptaulus and Metteniusa, opposite to alternate in Gonocaryum, pits mostly 5–7 μm in horizontal diameter, up to 10 μm in Pseudobotrys and 9–14 μm in Cardiopteris, nonvestured. Vessel-ray pits similar to intervessel pitting in shape and size, except for Metteniusa showing opposite to scalariform pits with reduced pit borders (pit borders 8–20 μm in horizontal diameter). Wall sculpturing absent. Tyloses absent. Tangential diameter of vessels (30–)40–260(–475) μm, vessel elements (300–) 470–2230(–2800) μm long. Tracheids present in Cardiopt- eris (Fig. 5E), (700–)1,050(–1,400) μm. Fibres nonseptate, fibres often thin- to thick-walled, thick-walled in Pseudo- botrys and very thick-walled in Gonocaryum ; pits conspicuously bordered and concentrated in tangential and radial walls (Fig. 5F), pit borders 5–9 μm in horizontal diameter; fibre length (1,175–)2,190–4,240(–4,900) μm. Axial parenchyma often diffuse-in-aggregates and scanty paratracheal (Fig. 5B), tendency to form short narrow bands in Dend- robangia, Gonocaryum and Metteniusa, in Cardiopteris lignified parenchyma concentrated around vessels (scanty and vasicentric) co-occuring with a nonlignified parenchymatous ground tissue (Fig. 5A); 3–11 cells per parenchyma strand. Uniseriate rays scarce to common, 0–11 rays/mm, but completely absent in Pseudobotrys dorae (Fig. 5F), (100–)330–1,070(–2,400) μm long, consisting of square to upright cells. Multiseriate rays narrow (2–4-seriate) in Car- diopteris and Leptaulus, up to 7-seriate in Citronella, up to
8-seriate in Dendrobangia, up to 9-seriate in Gonocaryum, and wider in Metteniusa (5–11-seriate) and Pseudobotrys (11–19-seriate, Fig. 5C), (300–)620–2,500(– > 6,000) μm high, 1–11 rays/mm, consisting of exclusively procumbent body ray cells in Cardiopteris, Leptaulus, a combination of procumbent and square body ray cells in Citronella, Den- drobangia, Gonocaryum, Metteniusa and Pseudobotrys and 1–6 rows of square to upright marginal ray cells; indistinct sheath cells in Citronella and Dendrobangia ; a small percentage of rays fused in Leptaulus. Dark amorphous contents in ray cells. Few prismatic crystals in body and marginal ray cells of Citronella (sometimes in chambered cells), Dendrobangia (often in chambered cells, Fig. 5G), and in nonchambered cells of Gonocaryum, Metteniusa and Pseudobotrys, few small styloids present in rays of Citronella, Dendrobangia, Gonocaryum and Leptau- lus, druses in nonlignified parenchyma of Cardiopteris, prismatic crystals also in chambered axial parenchyma cells of Dendrobangia, crystal sand absent. A specific type of cambial variants present in Cardiopteris moluccana : lignified islands consisting of vessels surrounded by paratracheal parenchyma, vasicentric tracheids (Fig. 5E) and fibres are embedded in nonlignified parenchymatous ground tissue (Fig. 5A).
Stemonuraceae (campanulids, Aquifoliales) (Cantleya 1/1, Codiocarpus 1/2, Discophora 1/2, Gastrolepis 1/1, Gomphandra 1/ca. 50, Hartleya 1/1, Lasianthera 2/2, Medusanthera 1/11, Stemonurus 2/ca. 20, Whitmorea 1/1; Fig. 6): Growth ring boundaries generally absent, indistinct in Hartleya and Stemonurus. Diffuse-porous. Ves- sels (4–)7–45(–50)/mm2, exclusively solitary in Cantleya and Lasianthera africana, solitary and in radial multiples and/or tangential multiples of 2–4 in most genera (Fig. 6A), additional vessel clusters of 3–7 cells in Stemonurus (Fig. 6B); vessel outline often slightly angular; perforation plates exclusively simple in Cantleya, predominantly simple ( > 90%) in Codiocarpus, Medusanthera (Fig. 6C) and Whitmorea, largely simple in Lasianthera (75% to > 90%), Stemonurus (75% to > 90%), Discophora (80%), and Gomphandra (75%), and simple/scalariform perforations equally abundant in Hartleya, scalariform perforations with (1–)3–18(–64) bars, occasionally double simple perforations in Discophora. Intervessel pits generally opposite to scalariform, although an alternate arrangement sometimes observed in Discophora and Lasi anthera, pits 5–35 μm in horizontal diameter, nonvestured. Vessel-ray pits similar to intervessel pits in shape and size, except for the scalariform vessel-ray pitting with reduced pit borders in species of Cantleya, Gastrolepis, Hartleya, Lasianthera, Medusanthera and Stemonurus, vessel-ray pits unilaterally compound in Medusanthera (Fig. 6D). Wall thickenings absent. Tyloses absent. Tangential diameter of vessels (30–)45–130(–190) μm, vessel elements (300–)600–1,510(–2,000) μm long. Tracheids absent. Fi-
539
bres nonseptate with distinctly bordered pits concentrated in tangential and radial walls, often thick- to very thick- walled (Fig. 6A–B) except in Gastrolepis and Gomphandra, (1,100–)2,000–3,650(–4,500) μm long, pit borders 4–9 μm in horizontal diameter. Axial parenchyma typically dif-
fuse-in-aggregates and scanty paratracheal, paratracheal parenchyma dominant in species of Gastrolepis, Lasian- thera, Stemonurus (Fig. 6B) and Whitmorea, tendency to banded axial parenchyma of 1–2 cells wide in Disco- phora (Fig. 6A), banded axial parenchyma prevailing in
Fig. 5. LM (A–C, G) and SEM micrographs (D–F) showing the wood anatomical diversity of Cardiopteridaceae. A, Cardiopt- eris moluccana, transverse section (TS), islands of lignified cells (vessels, tracheids, fibres and axial parenchyma) embedded in unlignified parenchymatous ground tissue (arrows); B, Citronella silvatica, TS, solitary vessels, diffuse-in-aggregates axial parenchyma; C, Pseudobotrys dorae, tangential longitudinal section (TLS), wide multiseriate rays, uniseriate rays are lacking; D, Gonocaryum crassifolium, radial longitudinal section (RLS), part of reticulate vessel perforation; E, Cardiopteris moluccana, TLS, tracheids with conspicuously bordered pits; F, Dendrobangia boliviana, TLS, fibre-tracheid with distinctly bordered pits; G, Dendrobangia boliviana, RLS, prismatic crystals in chambered body and marginal ray cells.
540
Medusanthera (1–3 cells wide) and Gomphandra (1–2 cells wide); 4–14 cells per parenchyma strand. Uniseriate rays generally scarce to even completely absent in Medu- santhera, 0–5 rays/mm, (50–)210–750(–1,550) μm high, consisting of square to procumbent ray cells. Multiseriate
rays generally 2–7-seriate, up to 9-seriate in Whitmorea and up to 10-seriate in Stemonurus, rays wider in Gom- phandra (10–17-seriate) and Medusanthera (9–11-seriate), (300–)835–2,810(–4,400) μm high in most genera, but much higher in Gomphandra (Fig. 6E) and Medusanthera,
Fig. 6. LM (A–B, E) and SEM (C–D, F) micrographs showing wood anatomical variation in Stemonuraceae. A, Discophora guianensis, transverse section (TS), vessels solitary and in short radial multiples, axial parenchyma diffuse-in-aggregates with tendency to short narrow bands, fibres very thick-walled; B, Stemonurus malaccensis, TS, vessels sometimes in clusters, axial parenchyma mainly paratracheal, fibres very thick-walled; C, Medusanthera laxiflora, radial longitudinal section (RLS), vessel perforations simple or scalariform; D, Medusanthera laxiflora, RLS, unilaterally compound vessel- ray pitting; E, Gomphandra luzoniensis, tangential longitudinal section (TLS), wide multiseriate rays; F, Lasianthera africana, RLS, crystal sand in ray cell.
541
2,500–9,000 μm high, multiseriate ray density 1–9 rays/ mm, rays consisting of procumbent body ray cells in Can- tleya, Codiocarpus, Stemonurus and Whitmorea, or a mixture of procumbent and square cells in the other genera, and 1–5(–10) rows of square to upright marginal ray cells; a small percentage of rays fused in Codiocarpus, Lasian- thera, Stemonurus and Whitmorea ; indistinct sheath cells sometimes present in Discophora, Gomphandra, Hart- leya, Lasianthera and Stemonurus; thick-walled sclereids in rays of Medusanthera. Dark amorphous contents in ray cells including crystal sand in Discophora and Lasian- thera. Prismatic crystals present in rays of Discophora, Gastrolepis, Lasianthera, Medusanthera, Stemonurus (also in fibres of S. umbellatus) and Whitmorea, few small styloid-like crystals present in nearly all genera (except for Gastrolepis and Lasianthera), clustered crystals present in Gastrolepis, Medusanthera, Stemonurus and Whitmo- rea, crystal sand present in Discophora, Gomphandra, Hartleya and Lasianthera (Fig. 6F), crystals absent from axial parenchyma. No cambial variants.
Pennantiaceae (campanulids, Apiales) (Pennantia 2/3; Fig. 7): Growth ring boundaries indistinct (Fig. 7B) to distinct. Diffuse-porous. Vessels (48–)70(–30)/mm2, predominantly solitary (Fig. 7A–B); vessel outline angular; perforation plates exclusively scalariform (Fig. 7C) with (18–)34(–56) bars, sometimes reticulate portions in P. cunninghamii. Intervessel pits opposite, pits 3–4 μm in horizontal diameter, nonvestured. Vessel-ray pitting comparable to intervessel pitting in shape and size. Helical thickenings present throughout vessel elements and in fibres in P. corymbosa (Fig. 7D), and sometimes in tails of vessel elements in P. cunninghamii. Tyloses absent. Tangential diameter of vessels (30–)55(–95) μm, vessel elements (700–)1,290(–2,250) μm long. Tracheids occasionally present, (1,400–)1,950(–2,700) μm long. Nonseptate fibres with distinctly bordered pits concentrated in tangential and radial walls, very thin-walled, (1,300–)2,030(–2,900) μm long, pit borders 4–6 μm in diameter. Axial parenchyma diffuse or diffuse-in-aggregates and scanty paratracheal; 5–10 cells per parenchyma strand. Uniseriate rays common, 5–9 rays/mm, consisting of square to upright cells, length (200–)500(–1,100) μm. Multiseriate rays often 4–8-seriate (Fig. 7E–F), (550–)1,200 (–2,000) μm high, 1–6 rays/mm, consisting of procumbent body ray cells with 1–4( > 20) rows of upright to square marginal ray cells; indistinct sheath cells sometimes present in P. cunninghamii (Fig. 7E). Dark amorphous contents absent in rays. All types of mineral inclusions absent. No cambial variants.
Phylogenetic analyses based on maximum parsimony. — The maximum parsimony analysis of the molecular data finds 8,190 most parsimonious trees with length 15,756 (CI 0.278; RI 0.523), and generally lacks bootstrap support (BS) for the relationships between the
major lineages of Icacinaceae s.str. (Fig. 8A). Regarding Icacinaceae s.str., the following groups are well supported within lamiids: Emmotum-group (BS 100; represented by only two out of six genera), Apodytes-group (BS 100), which is weakly supported as sister to Oncotheca (BS 58), and Icacina-group (BS 100; represented by 11 out of 25 genera). Other lineages within lamiids include Gar- ryales (BS 99; including Eucommia), and a large clade consisting of Lamiales-Solanales-Gentianales plus Vahlia (Vahliaceae) and Borago-Hydrophyllum (Boraginaceae; BS 100), which are unplaced as to order. Within campanulids, Aquifoliales is the earliest diverging clade, including the monophyletic Cardiopteridaceae and Stemonuraceae (BS 82 and 98, respectively), which show a strong sister relationship (BS 100). Both families are in turn sister to the rest of Aquifoliales (BS 98). Aquifoliales is sister to a large clade (BS 100) comprising Asterales (BS 100), Apiales (BS 100), Dipsacales (BS 100), and Berzelia-Escallonia (BS 98). Apiales and Dipsacales are sisters (BS 80). The two species of Pennantia are strongly nested at the base of Apiales (BS 100).
The maximum parsimony analysis of the combined matrix reveals 120 most parsimonious trees with length 16,095.9 (CI 0.274; RI 0.521). The topology shows more resolution within the Icacinaceae s.str. relationships (Fig. 8B), although BS is often low to moderate. Within lamiids, the earliest diverging lineage is formed by a moderately supported monophyletic group (BS 72), including two poorly supported subclades: subclade 1 (BS 63) comprises Cassinopsis, which is sister to the Emmotum-group; and subclade 2 (BS 59) comprises Oncotheca, which is sister to the Apodytes-group. The following group that branches off includes Garryales (BS 99), which is very weakly supported (BS 51) as sister to the rest of the lamiids. Next, the Icacina-group (BS 100) is weakly supported (BS 59) as sister to a large clade (BS 100) comprising Lamiales- Solanales-Gentianales plus Vahlia and Borago-Hydro- phyllum. Within campanulids, relationships and bootstrap supports are very similar to the molecular analysis, except for Berzelia-Escallonia, which are weakly supported (BS 53) as sister to Asterales. If the three MorphoCode characters are omitted from the analysis, the topology and BS remain similar.
Phylogenetic analyses based on Bayesian inference. — The Bayesian analysis of the molecular matrix (Fig. 9A) is generally better resolved than Fig. 8A, but posterior probabilities (PP) between different lineages of Icacinaceae s.str. are often too low to make conclusive comments on their relationships. In Fig. 9A, the first diverging evolutionary lineage of lamiids is formed by a weakly supported clade (PP 0.76) and is divided into two subclades: subclade 1 (PP 0.81) includes two branches, one (PP 0.69) including the Emmotum-group (PP 1.00), which is sister to Cassinopsis, and the other (PP 0.88)
542
comprising the Apodytes-group (PP 1.00) and Oncotheca ; subclade 2 includes Garryales (PP 1.00). The Icacina- group (PP 1.00) is the next lineage that diverges within lamiids (PP 0.87), followed by Gentianales, Solanales plus Borago-Hydrophyllum (PP 0.91), Vahlia, and Lamiales (PP 1.00). Within campanulids, the first diverging lineage is Aquifoliales (PP 1.00) including the well-supported
sister families Cardiopteridaceae (PP 1.00) and Stemonu- raceae (1.00), which are in turn sister to the rest of the order. Subsequent lineages include Asterales, which are sister to a clade formed by Berzelia-Escallonia (PP 1.00), Apiales (PP 1.00; including Pennantia at the base) and Dipsacales.
The Bayesian analysis of the combined data (Fig. 9B)
Fig. 7. Wood anatomical variation in Pennantiaceae based on LM (A–B, E–F) and SEM (C–D). A, Pennantia cunninghamii, transverse section (TS), mainly solitary vessels and thin-walled fibres; B, Pennantia corymbosa, TS, mainly solitary vessels and thin-walled fibres; C, Pennantia cunninghamii, radial longitudinal section (RLS), scalariform perforation plate; D, Pen- nantia corymbosa, RLS, helical thickenings in inside walls of vessels; E, Pennantia cunninghamii, tangential longitudinal section (TLS), multiseriate rays with long uniseriate ends (arrow); F, Pennantia corymbosa, TLS, uni- and multiseriate rays.
543
Fig. 8. Strict consensus trees of the maximum parsimony analyses based on molecular data (A) and combined wood anatomical-molecular data (B). Bootstrap values are given above the branches. Taxa written in bold represent Icacinaceae s.l.; genera marked with “+” refer to taxa with climbing members. APO, Apodytes-group of Icacinaceae s.str.; CAR, Car- diopteridaceae; CAS, Cassinopsis; EMM, Emmotum-group of Icacinaceae s.str.; ICA, Icacina-group of Icacinaceae s.str.; PEN, Pennantiaceae; STE, Stemonuraceae.
544
Fig. 9. Bayesian phylogenies based on the molecular data (A) and the combined anatomical-molecular matrix (B). Poste- rior probabilities (PP) are indicated as asterisks above the branches: one asterisk refers to PP-values between 0.95 and 0.99, two asterisks stand for maximum PP-values of 1.00; branches without asterisks have PP-values below 0.95. Taxa written in bold represent Icacinaceae s.l.; genera marked with “+” refer to taxa with climbing members. APO, Apodytes- group of Icacinaceae s.str.; CAR, Cardiopteridaceae; CAS, Cassinopsis; EMM, Emmotum-group of Icacinaceae s.str.; ICA, Icacina-group of Icacinaceae s.str.; PEN, Pennantiaceae; STE, Stemonuraceae.
545
breau-Callen, 1973) also supports the isolated position of Sarcostigma as compared to the other lianas.
From a wood anatomical point of view, the remaining lineages within Icacinaceae s.str., i.e., Cassinopsis, Apodytes-group (without Rhaphiostylis) and Emmotum- group, are very similar and easily distinguishable from the Icacina-group due to a so-called primitive set of wood features in the Baileyan sense (Bailey & Tupper, 1918; Kribs, 1937). This combination of wood features, which is often encountered in basal angiosperm groups, includes: (1) exclusively scalariform perforations with usually many bars (range of means 8–80; Fig. 1E), (2) mainly opposite (to scalariform) vessel pitting (Fig. 1F), (3) very long vessel elements and fibres (range of means 840–2,220 μm and 1,680–3,120 μm, respectively), and (4) mainly diffuse- in-aggregates and scanty paratracheal axial parenchyma (Fig. 1A–B). Within the Emmotum-group, Emmotum and Ottoschulzia can be easily recognized by their scalariform perforations with less than 11 bars. The lianescent genus Rhaphiostylis does not share the characters just mentioned, but perfectly fits with the anatomy of the Icacina-group (except for the banded parenchyma).
The wood anatomical diversity within Cardiopteri- daceae is more variable, especially in perforation plate morphology (exclusively scalariform, scalariform/simple, exclusively simple, and occasionally reticulate; Fig. 5C), axial parenchyma (diffuse-in-aggregates, banded and scanty paratracheal; Fig. 5B) and ray characters (large differences in width and height). If the climbing Cardi- opteris is not considered, more or less stable wood characters are predominantly solitary vessels, mainly opposite (to slightly alternate) vessel pitting, and very long vessel elements and fibres (800–2,230 μm and 2,190–4,240 μm, respectively), but this set of wood characters is widely scattered within angiosperms. Cardiopteris can be defined by large circular vessel pits (9–14 μm vs. 5–8 μm in the other genera), narrow (2-seriate) and low ( < 1 mm) multiseriate rays, and especially by the nonlignified parenchymatous ground tissue containing lignified islands of xylem (represented by vessels, fibres, vasicentric tracheids and paratracheal parenchyma; Fig. 5A). The wood anatomy of the controversial genus Metteniusa is similar to the other genera of Cardiopteridaceae, but Pseudobot- rys (the other genus that was placed in the family with hesitation) is strikingly different from all other taxa studied due to its deviating ray structure (no uniseriate rays, very wide multiseriate rays, Fig. 5C).
Stemonuraceae resemble Cardiopteridaceae in several aspects, amongst others the opposite to alternate intervessel pitting and the considerable length of vessel elements and fibres (mean ranges 600–1,510 μm and 2,000–3,650 μm, respectively). Nevertheless, Stemonuraceae exhibit a more homogeneous wood structure than Cardiopteri- daceae, which can be demonstrated by: (1) a considerably
results in a topology similar to that in Fig. 9A, notwith- standing some minor differences. For instance, the earliest diverging lineage within lamiids remains the same, although its support decreases from PP 0.76 to 0.53, and two well-supported subclades are present: subclade 1 (PP 0.99) includes the Apodytes-group (PP 1.00), which is sister to another group comprising the Emmotum-group, Cassin- opsis and Oncotheca, although support for this sister relationship is almost lacking (PP 0.52); subclade 2 includes Garryales (PP 1.00). In the rest clade of lamiids (PP 1.00), the earliest diverging clade is formed by the Icacina-group with strong support (PP 1.00); the other large clade includes Lamiales-Solanales-Gentianales plus Vahlia and Borago-Hydrophyllum (PP 1.00). The relationships within campanulids are nearly identical to Fig. 9A.
DISCUSSION Diagnostic wood characters of Icacinaceae
s.str., Cardiopteridaceae, Stemonuraceae and Pen- nantiaceae. — The wood anatomical diversity of the distinctly polyphyletic Icacinaceae is huge (Figs. 1–7), as already mentioned by Bailey & Howard (1941a–d). Never- theless, most of the diversity observed is due to differences in habit (lianas vs. erect trees and shrubs), while the wood anatomy of the climbing and especially the nonclimbing taxa separately is rather homogeneous. Consequently, the wood structure of the four currently circumscribed segregate families does not offer straightforward characters to define the family boundaries sensu Kårehed (2001). This is not surprising, because (1) Icacinaceae s.str. are probably not monophyletic (see also Figs. 8–9), and (2) Cardiopteridaceae and Stemonuraceae are strongly supported as sisters. Nevertheless, some wood features merit special emphasis because of their predictive value to assign a species to one of the four families.
Within Icacinaceae s.str., the Icacina-group (including the bulk of the climbing genera plus some nonclimbing ones) is rather homogeneous and can be easily distinguished from the other Icacinaceae lineages due to a combination of anatomical characters. These include simple vessel perforations, solitary vessels plus tangential multiples, a tendency to alternate vessel pitting, relatively short vessel elements and fibres (often between 300–700 μm and 600–1,000 μm respectively), banded and vasicentric axial parenchyma, high and wide multiseriate rays, and the occurrence of cambial variants. Sarcostigma has a peculiar anatomy within the Icacina-group because of: (1) interxylary phloem (Fig. 3A), (2) few septate libriform fibres that co-occur with the normal, distinctly bordered fibres (Fig. 2D), and (3) scarce, narrow (2-seriate) and low ( < 1 mm) multiseriate rays. Evidence from leaf anatomy (Van Staveren & Baas, 1973) and pollen morphology (Lo-
546
higher percentage of vessel multiples (Fig. 6A), (2) mainly simple perforations that co-occur with few scalariform ones having on average 9 bars (Fig. 6C), (3) scalariform to opposite vessel pitting, (4) very thick-walled fibres (Fig. 6A–B), and (5) the tendency to form more para tracheal than apotracheal parenchyma (Fig. 6B). This set of characters is confirmed by additional observations in extra species of Gomphandra and Lasianthera from the Kew collection (pers. obs.). Furthermore, about half of the Stemonuraceae species have scalariform vessel-ray pits with strongly reduced pit borders (also present in Cas- sinopsis, Metteniusa and Poraqueiba ; Fig. 1F), and the genera Discophora, Gomphandra, Hartleya and Lasian- thera are the only ones that contain crystal sand in their wood rays (Fig. 6F; in contrast to Heintzelman & Howard (1948) who mentioned that crystal sand is always absent in their wood samples). Crystal sand outside the wood has also been observed in three other Stemonuraceae genera (Cantleya, Medusanthera, Stemonurus), but despite the phylogenetic significance of this character in many groups (e.g., Sapotaceae; pers. obs.), crystal sand outside the wood structure is also present in three genera of Cardiopteri- daceae and six genera of Icacinaceae s.str. (Heintzelman & Howard, 1948).
The two species of Pennantia studied resemble very much the so-called primitive complex of wood characters that is also typical of the nonclimbing lineages of Icacinaceae s.str. and some Cardiopteridaceae (Citronella, Dendrobangia, Metteniusa, Pseudobotrys) in having mainly solitary vessels (Figs. 7A–B), exclusively scalariform perforations (Fig. 7C), predominantly opposite vessel pitting, fibres with distinctly bordered pits, and mainly diffuse-in-aggregates parenchyma that co-occur with scanty paratracheal parenchyma. Nevertheless, two wood features can be used to define Pennantia from the traditionally defined Icacinaceae, i.e., minute to small vessel pitting (3–4 μm) and the absence of prismatic crystals. In addition, the combination of width (4–8-seriate) and height (on average 1,000–1,400 μm) of the multiseriate rays in Pennantia usually offers additional distinguishing characters (Table 2).
Anatomy of climbers versus nonclimbers. — As mentioned before, the enormous wood anatomical diversity in the secondary xylem is largely attributed to differences in habit (lianas vs. erect trees and shrubs). The climbing group, represented by the taxa flanked with “+” in Figs. 8–9, can be characterised by mainly solitary vessels co-occurring with a small percentage of tangential multiples (Fig. 1C–D), exclusively or predominantly ( > 95%) simple perforations, short to medium-sized vessel elements (range of means 175–880 μm), opposite to alternate vessel pitting, short to medium-sized fibres (range of means 610–1,540 μm), an abundance of banded and vasicentric axial parenchyma (Fig. 1D), high multiseri-
ate rays (often more than 10 mm in length) with entirely or partly unlignified ray cells, and cambial variants (Fig. 3A–E). The nonclimbing group, represented by the bold genus names that are not flanked with “+” in Figs. 8–9, can easily be distinguished from the climbing group based on the perforation plate type (exclusively scalariform with often many bars, range of means 8–80, or a mixed occurrence of simple and scalariform perforations), vessel element length (typically long, but short in Rhyticaryum; range of means 500–2,230 μm), opposite or opposite to scalariform vessel pitting, diffuse-in-aggregates (to nar- rowly banded) and scanty paratracheal axial parenchyma, and long fibres (range of means 970–4,240 μm).
Due to the mixed occurrence of lianescent and erect genera in Icacinaceae s.str. and Cardiopteridaceae, the lianescent habit within these two families blurs the phylogenetic signal in the wood structure. For example, both members of the Apodytes-group (Rhaphiostylis lianas and Apodytes trees) have a completely different wood anatomy. Likewise, the wood structure of the only climbing genus of Cardiopteridaceae (Cardiopteris) is unique within the family, and even strongly deviates from the other lianescent genera of Icacinaceae s.str. (see next section). On the other hand, the wood structure of the few erect representatives in the Icacina-group (e.g., Desmostachys, Merrillioden- dron, Rhyticaryum) reveals some interesting similarities compared to the common climbing Icacina-group members, such as exclusively simple perforations, a strong tendency to form alternate vessel pits, relatively short vessel elements (range of means 450–730 μm) and fibres (range of means 970–1,630 μm), and banded and/or vasicentric axial parenchyma. This means that these similarities are independent from differences in habit, indicating that this group may not be closely related with other Icacinaceae s.str. as shown by our phylogenetic analyses (Figs. 8–9).
Cambial variants. — Apart from the relatively young stems studied of Mappianthus and Polyporandra, all climbing taxa studied exhibit stems with cambial variants, which can provisionally be divided into four major types: (1) interxylary phloem (Fig. 3A), (2) nonlignified parenchymatous ground tissue with islands of lignified xylem (Fig. 5A), (3) successive cambia (Fig. 3B–E), and (4) furrowed xylem with elongated phloem zones (Fig. 4A–B). Interxylary phloem, irregularly scattered as islands in the wood cylinder, is typical of Sarcostigma (Fig. 3A; cf. Bailey & Howard, 1941a; Metcalfe & Chalk (1950) also reported successive cambia in Sarcostigma, but original slides and slides from four additional species in the Bai- ley & Howard collection only show interxylary phloem). The Bailey & Howard collection also demonstrates that the occurrence of interxylary phloem is not restricted to Sarcostigma: Pleurisanthes flava also has interxylary phloem islands that are more or less arranged in rings. The second type of cambial variants, represented by non-
547
lignified ground tissue with islands of lignified xylem, is observed only in Cardiopteris (Fig. 5A; cf. Bamber & ter Welle, 1994). Another distinguishing anatomical character of Cardiopteris found in the literature is the presence of laticifers in the pith, bark and leaves (Thouvenin, 1891). However, laticifers in pith and bark are not present in our material of C. moluccana. It remains unknown whether all remaining climbing genera are characterised by successive cambia or not, because mature wood samples from many genera have not been investigated by us, nor in the literature due to lack of material. Furthermore, it seems that the presence of successive cambia may vary within a genus: Obaton (1960) observed a mature stem of 90 mm of Pyrenacantha mangenotiana that only exemplifies furrowed xylem with elongated phloem zones, while our study reveals elongated phloem regions as well as successive cambia in P. malvifolia, which is very similar to Phytocrene (Sleumer, 1942a; Metcalfe & Chalk, 1983; Fig. 3D). It can be suggested that all (except Sarcostigma and Pleurisanthes ?) climbing genera within the Icacina-group have successive cambia because of several reasons: (1) the predictive value of successive cambia in different groups (for instance in Caryophyllales, Jansen & al., 2000; Car- lquist, 2001), (2) the hypothesis that the Icacina-group is monophyletic (Figs. 8–9), (3) our observation that successive cambia are present in Iodes (a feature not mentioned in the literature due to lack of mature material), and (4) the recent observation of successive cambia in the new genus Sleumeria (Utteridge & al., 2005). If this should be true, two major types within the successive cambia- group could be distinguished based on the presence/ absence of elongated phloem zones that are developed by the (unilaterally-active portion of the vascular) cambium at places opposite to the detachment of the leaves (Bailey & Howard, 1941a; Sleumer, 1942a; Obaton, 1960). Gen- era that show the cambial variant type with the peculiar phloem formation all belong to the former Phytocreneae (Chlamydocarya, Fig. 4B; Miquelia, Phytocrene, Fig. 3D; Polycephalium, Pyrenacantha, Fig. 4A; Stachyanthus), and it has largely been observed in young stems (confirmed by the original slides of Bailey & Howard; mature material only available from Phytocrene and Pyrenacan- tha). According to Carlquist (2001), the presence of elongated phloem regions in co-occurrence with successive cambia might be a unique feature among flowering plants. The young stems of Chlamydocarya thomsonia (Fig. 4B) and Miquelia caudata (slide collection Bailey & Howard) are remarkable in this regard, because they develop successive cambia that are restricted to the elongated phloem zones (cf. Sleumer, 1942a).
Taxonomic position of Icacinaceae s.l. — Previ- ous analyses that dealt with the taxonomic position of Icacinaceae s.str. found that the family is nested at the base of lamiids, and may be closely allied with Garryales and
Oncotheca (Savolainen & al., 2000; Kårehed, 2001; Ste- vens, 2001 onwards; Bremer & al., 2002; Cameron 2003), while the interfamily relationships of Cardiopteridaceae and Stemonuraceae (both Aquifoliales) and Pennantiaceae (Apiales) have been established with much greater confi- dence (Kårehed, 2001, 2003; Figs. 8–9). Our maximum parsimony and Bayesian analyses of both molecular and combined data strongly support the position of the latter three families, but the relationships between the distinct lineages of Icacinaceae s.str. remain enigmatic. We have found further support that the family Icacinaceae s.str. does not form a monophyletic entity, an idea that has already been proposed by Kårehed (2001). Although many more representatives should be studied to comment more thoroughly on the relationships within Icacinaceae s.str., our analyses (Figs. 8–9) add some new insights: (1) the Apodytes-group, Emmotum-group and Cassinopsis form together with the genus Oncotheca (Oncothecaceae) a monophyletic group with moderate to strong support based on the combined maximum parsimony and Bayes- ian analyses (Figs. 8B, 9B); this clade is nested at the base of the lamiids and might be closely related with Garryales, (2) the Icacina-group seems well supported and may not be directly related with other lineages of Icacinaceae s.str.; a possible sister relationship with a clade represented by Gentianales-Solanales-Lamiales plus Boraginaceae and Vahliaceae is plausible, especially based on the Bayesian analysis of the combined data (Fig. 9B).
Wood anatomical comparison within the asterids. — As discussed before, the four possible (nonclimbing) candidates that may form the earliest diverging lineage within lamiids together with the Apodytes group, the Emmotum-group and Cassinopsis are Aucuba, Eucommia and Garrya (the only three genera in Gar- ryales) and Oncotheca (Oncothecaceae). From a wood anatomical point of view, these three genera perfectly fit within this early diverging lamiid lineage because of the joint occurrence of: (1) predominantly solitary vessels, (2) exclusively scalariform perforations, (3) nonseptate fibres with distinctly bordered pits (simple to minutely bordered pits and sometimes septate in Aucuba), and (4) mainly diffuse to diffuse-in-agregates axial parenchyma (Baas, 1975; Carpenter & Dickison, 1976; Dickison, 1982; Noshiro & Baas, 1998; original observations). These four characters are part of the so-called primitive Baileyan syndrome, which is in agreement with other basal lineages in asterids (see further). When the wood structure of Gar- ryales is compared with the Icacina-group

Documents

The wood anatomy of the polyphyletic Icacinaceae sl, and their relationships within asterids